Sphingolipid biosynthesis in Physcomitrella patens

(1)

Physcomitrella patens

Dissertation

for the award of the degree

“Doctor rerum naturalium”

of the University of Goettingen

within the GGNB doctoral program

“Microbiology and Biochemistry”

of the Georg-August University School of Science (GAUSS)

submitted by

JasminGömann

born in Holzminden

Göttingen 2020

(2)

II

Department for Plant Biochemistry, Albrecht-von-Haller-Institute for Plant Science, University of Goettingen

Prof. Dr. Volker Lipka

Department of Plant Cell Biology, Albrecht-von-Haller Institute for Plant Science, University of Goettingen

Prof. Dr. Andrea Polle

Department for Forest Botany and Tree Physiology, Buesgen-Institute, University of Goettingen

Members of the Examination Board Referee: Prof. Dr. Ivo Feußner

2^nd Referee: Prof. Dr. Volker Lipka

Department of Plant Cell Biology, Albrecht-von-Haller Institute for Plant Sciences, University of Goettingen

Further members of the Examination Board Prof. Dr. Andrea Polle

Department for Forest Botany and Tree Physiology, Buesgen-Institute, University of Goettingen

Prof. Dr. Stefanie Pöggeler

Department for Genetics of Eukaryotic Microorganisms, Institute for Microbiology and Genetics, University of Goettingen

PD Dr. Till Ischebeck

Jun.-Prof. Dr. Jan de Vries

Department of Bioinformatics (IMG), Institute for Microbiology and Genetics, University of Goettingen

Date of oral examination: November 16^th, 2020

(3)

III

“ ‘Tell me, though, Miss Whittaker, what is it that you admire in mosses?’. ‘Their dignity,’

Alma replied without hesitation. ‘Also, their silence and intelligence. I like that—as a point of study—they are fresh. They are not like other bigger or more important plants, which have all been pondered and poked at by hordes of botanists already. I suppose I admire their modesty, as well. Mosses hold their beauty in elegant reserve. By comparison to mosses, everything else in the botanical world can seem so blunt and obvious. Do you understand what I am saying? Do you know how the bigger, showier flowers can look at times like dumb, drooling fools—the way they bob about with their mouths agape, appearing so stunned and helpless?’ […] ‘Somebody must defend them, Mr. Pike! For they have been so overlooked, and they have such a noble character! In fact, I find the miniature world to be a gift of disguised greatness, and therefore an honor to study.’ “

- Elizabeth Gilbert, The Signature of All Things

(S.201, Bloomsbury Publishing. Kindle-Version)

(4)

IV

I hereby confirm that I wrote this dissertation entitled “Sphingolipid biosynthesis in Physcomitrella patens” on my own. No other sources and aids than quoted were used.

___________________________

Jasmin Gömann Göttingen 2020

(5)

V

1 Abstract

The complex sphingolipid classes glycosylceramides (GlcCers) and glycosyl inositolphosphorylceramides (GIPCs) are essential membrane components in plant cells.

However, the regulation of their synthesis and their distinct physiological roles in plants is still poorly understood. GlcCers and GIPCs both contain a ceramide backbone consisting of a long-chain base (LCB) that is connected to a fatty acid. The syntheses of these two complex sphingolipids are alternative pathways in plant metabolism. It is assumed that distinct structural modifications in the LCB moiety determine the metabolic fate and physiological function of sphingolipids. In the bryophyte Physcomitrella patens, channelling of sphingolipid metabolites into complex sphingolipid formation appears to be stricter than in vascular plants. The physiological relevance of GlcCers, GIPCs, and their specific LCB moieties was therefore investigated in P. patens. P. patens GlcCers are enriched in ceramides with a dihydroxy, Δ4,8-diunsaturated LCB moiety while P. patens GIPCs mostly contain ceramides with a trihydroxy LCB moiety. The establishment of a sophisticated cultivation system and of various mutant characterisation assays is a prerequisite for in- depth examinations of P. patens mutants. P. patens knockout mutants were generated by homologous recombination that targeted key steps of sphingolipid biosynthesis. Disruption of the LCB C-4 hydroxylase, PpS4H, which is involved in GIPC formation, resulted in plants that were severely impaired in growth and development. These growth impairments might have derived from cell plate formation defects during cytokinesis. Loss of the trihydroxy LCB moiety also caused global changes in all sphingolipid classes. Disruption of the LCB Δ4- desaturase, PpSD4D, did not substantially affect plant viability. The mutant only showed mild cell elongation defects. However, sd4d-1 mutants had substantially reduced GlcCer levels, which confirms that LCB Δ4-desaturation is important for channeling sphingolipids into GlcCer formation in P. patens. In contrast, P. patens plants that had a disturbed glycosylceramide synthase, PpGCS, activity, were affected in plant growth and cell differentiation and showed cell death-like lesions. gcs-1 plants lacked all GlcCers and accumulated precursor hydroxyceramides. Cumulative findings from this work show that disruption of individual steps in P. patens sphingolipid biosynthesis differently affect plant physiology. The results give first insights into sphingolipid biosynthesis in P.patens. While some aspects of plant sphingolipid metabolism known from studies in Arabidopsis thaliana have been confirmed in the bryophyte, novel features of the sphingolipid pathway could also be identified in P. patens. These new findings contribute to our knowledge on how sphingolipid synthesis and function have diversified during land plant evolution.

(8)

2

2 Introduction

2.1 Structure and function of biological membranes

Biological membranes are natural barriers that separate the inside of a cell from its external environment. They also act as an interface between the intracellular and extracellular space and therefore enable the exchange of nutrients and information. Intracellular membranes further restrict individual organelles and thereby establish compartmentalisation within a eukaryotic cell. The spatial restriction of subcellular compartments enables the simultaneous performance of numerous processes in close proximity to each other. Cell viability relies on the orchestration of these different physiological processes. Biological membranes play pivotal roles in synchronising cellular processes. The major constituent of biological membranes is a phospholipid bilayer. Phospholipids contain one or two hydrophobic hydrocarbon chains that are connected to a hydrophilic head group. The combination of hydrophilic and hydrophobic building blocks defines lipids as amphipathic molecules which is a crucial feature for membrane bilayer formation. The polar head group is oriented towards the surrounding aqueous phase, while the non-polar hydrocarbon chains face each other. They thereby form the inner and outer leaflets of biological membranes. The inner leaflet faces the cytosol of a cell or the lumen of an organelle while the outer leaflet faces the extracellular space or in case of plant cells, the apoplast.

The plasma membrane (PM) is a semipermeable lipid bilayer that defines the boundary of a cell. It separates the intracellular space from the extracellular environment.

In plant cells the PM more accurately defines the symplast of a cell. The plant PM is additionally attached to a surrounding cell wall. The entirety of the intercellular space and all cell walls is called the apoplast. The PM is composed mainly of phosphoglycerolipids, sterols, sphingolipids, membrane proteins, and carbohydrates that are attached to some of the lipids and proteins at the exterior surface (Mamode Cassim et al., 2019). At the cytoplasmic side the PM is connected to the cytoskeletal network and therefore offers structural support of the cell (Sackmann, 1990). Through incorporated ion channels and surface proteins the PM regulates the import and export of nutrients, metabolites, and signalling compounds and therefore has a vital role in the perception and transduction of incoming information.

Membrane lipids are not equally distributed between the two monolayers of the PM which results in lipid asymmetry. While phophatidylcholine (PC), sterols, and sphingolipids are prevalent lipids in the apoplastic outer monolayer of the PM, other unsaturated phospholipids like phosphatidylethanolamine (PE), phosphatidylinositol (PI) and phosphatidylserine (PS) are more abundant in the cytosolic inner monolayer (Devaux &

Morris, 2004; Tjellström et al., 2010; Cacas et al., 2016; Mamode Cassim et al., 2019). This

(9)

3 unequal distribution of membrane lipids equips each PM leaflet with distinct biophysical properties. The two monolayers have differing charges which together with incorporated ion channels contribute the establishment of the membrane potential. Furthermore, sugar residues and glycosylphosphatidylinositol- (GPI) anchored proteins at the outer monolayer serve in cell and pathogen recognition and hence in the activation of downstream signalling cascades (Borner et al., 2005; Lenarčič et al., 2017).

Plants are sessile organisms and are therefore restricted in their abilities to protect themselves against unfavourable environmental conditions. Therefore, they had to develop different adaptation strategies to surrounding putative threats. A major strategy is the dynamic adjustment of membrane properties. For instance, PM lipid composition can be adjusted to maintain membrane fluidity in changing temperatures, which increases plant tolerance of cold stress (Miquel et al., 1993; Uemura et al., 1995). Surface proteins embedded into the PM bilayer perceive pathogen components such as pathogen associated molecular patterns (PAMPs) and microbe-associated molecular patterns (MAMPs) which in turn initiates a signal transduction cascade that activates the plant’s immune response (Gómez-Gómez & Boller, 2000; Sanabria et al., 2010). The dynamic short-term adjustment of membrane properties underlies strict control mechanisms that mediate membrane organisation.

2.2 The evolution of membrane organisation models

The fluid-mosaic model proposed by Singer and Nicolson in 1972 describes a membrane as a heterogenous, fluid phospholipid bilayer into which membrane proteins are randomly incorporated in a mosaic-like pattern (Singer & Nicolson, 1972). The fluid-mosaic model is the first model that described the membrane as a dynamic compartment and has been a widely accepted concept of PM organisation. Following studies concerning lipid trafficking and membrane-associated signal transduction suggested that proteins may not be randomly distributed within membranes (van Meer & Simons, 1982; Lisanti & Rodriguez- Boulan, 1990). Based on these and other studies, the fluid mosaic model had been adjusted over the past decades. However, the basic principle has been maintained.

In 1997 Simons and co-workers proposed a new model of PM organisation, which is known as the ‘lipid raft hypothesis’ (Simons & Ikonen, 1997). The model describes the presence of micro- and nanoscale, temporary membrane regions, or ‘lipid rafts’, that are enriched in sterols, sphingolipids, and certain proteins (Simons & Vaz, 2004). According to this new concept, membrane proteins are distributed in these lateral lipid partitions instead of being randomly dispersed in the membrane. The proposed function of the sterol- and sphingolipid-rich domains is the lateral segregation and diffusion of membrane components,

(10)

4

especially of proteins. This becomes most important during membrane-associated signalling cascades that are induced by incoming stimuli (Simons & Toomre, 2000).

In lipid bilayers with high levels of sterols the co-existence of two liquid lipid phases is observed (Recktenwald & McConnell, 1981). Co-existence of distinct liquid phases is enabled by lateral phase separation. This means that different phases are laterally separated within the plane of the membrane. Two liquid crystalline lamellar phases are described: the liquid-ordered (Lo) phase and the liquid-disordered (Ld) phase (Ipsen et al., 1987; Scheiffele et al., 1997). The Lo phase is characterised by high levels of sterols, sphingolipids, and saturated phospholipids. Because of the high content of saturated lipids and the intercalation of sterols, Lo phases are more rigid due to tight lipid packaging and are therefore considered more ‘ordered’. The Ld phase has higher levels of unsaturated phospholipids. Lipids in the Ld phase can diffuse and rotate more freely within the bilayer plane and the phase is therefore more fluid, or ‘disordered’. Membrane domains can have either a Lo or a Ld phase-like structure. Lipid rafts have high levels of sterols and sphingolipids and are therefore considered to be in the Lo phase (Scheiffele et al., 1997).

The term ‘membrane domain’ is hence a more general description for regions with distinct lipid and protein composition, while ‘lipid rafts’ are a subtype of membrane domains.

According to the raft hypothesis, membrane domains with distinct phase structures are considered to control membrane protein clustering.

Experimental evidence for the raft hypothesis has been provided by lipid purification studies using detergents. Detergents have a conical, amphiphilic molecular structure which causes spontaneous micelle formation in aqueous solutions. These compounds are therefore referred to as ‘curvophilic’. Phospholipids, however, form lipid bilayers and are therefore referred to as ‘curvophobic’ (Lichtenberg et al., 2005). High detergent concentrations cause membrane solubilisation with phospholipids residing in detergent micelles. Membrane fractions with different lipid compositions are solubilised at different lipid/detergent ratios (Lichtenberg, 1985). This aspect has been used as a biochemical tool in studying the composition of biological membranes. Some membrane fractions are highly resistant against detergent solubilisation and stay in the lipid bilayer conformation at even high detergent levels. Lipid rafts that contain mixtures of sphingolipids, sterols, and GPI- anchored proteins are described as highly resistant to detergent solubilisation (Hanada et al., 1995). Therefore, detergent-resistant membranes (DRMs) have often been identified as in vitro versions of lipid rafts. However, this association might be misleading since the experimental extraction procedure might induce artificial DRM formation which might not be found in native membranes (Lichtenberg et al., 2005). This controversy caused many scientific debates over the existence of lipid rafts in biological membranes and called for new methods to investigate membrane lipid heterogeneity. Over the years, new

(11)

5 technologies including advanced lipid analytics, proteomics, and methods to visualise lipid raft markers in vivo have provided more evidence for lipid clustering in cell membranes (Pike, 2009; Cacas et al., 2016). These studies have supported the lipid raft hypothesis.

Especially investigations of model membranes that mimic the lipid composition of native membranes have been useful tools in improving our understanding of membrane organisation (Wesołowska et al., 2009; Lin & London, 2014; Grosjean et al., 2015; Grosjean et al., 2018). Most membrane studies have been conducted on animal systems. However, animal membranes have different lipid compositions compared to plant membranes. In the mammalian PM cholesterol, sphingomyelin, and glycosphingolipids are the most abundant sterol and sphingolipid compounds. Hence, lipid rafts are considered to be mainly formed by the interaction between these compounds (Simons & Ikonen, 1997).

2.3 Plant lipid diversity

Plants contain a variety of lipids with different head group and backbone compositions (Table 1). These structurally varying lipid classes have different cellular distribution patterns and confer a multitude of physiological functions in the plant cell. Triacylglycerols (TAGs) are located in lipid droplets and are enriched in seeds where they serve as high-energy storage compounds (Xu & Shanklin, 2016). Diacylglycerols (DAGs) were described to act in lateral root development under mild salt stress in A. thaliana and may therefore play a role as second messengers in plants (Peters et al., 2014). DAGs are further assumed to be important building blocks of plant cell membranes where they induce a negative curvature, which might be an important feature during membrane fusion events (Szule et al., 2002).

The majority of lipids have a function as structural elements in various membranes.

Galactolipids such as monogalactosyldiacylglycerols (MGDGs) and digalactosyldiacylglycerols (DGDGs) represent highly abundant plant lipids that are enriched in thylakoid membranes where they have crucial roles in maintaining the integrity of photosynthetic active membranes (Dorne et al., 1990; Joyard et al., 1998; Dörmann &

Benning, 2002). DGDGs are also found in minor amounts in the plasma membrane and are specifically enriched upon phosphate deprivation (Andersson et al., 2003; Andersson et al., 2005). Membranes of the endoplasmic reiticulum (ER), the Golgi apparatus, and the tonoplast are part of the secretory pathway and make up a big proportion of the lipid content of a plant cell. The organelle membranes contain large amounts of phosphoglycerollipids and minor amounts of sterols and sphingolipids. A gradient of sterols and sphingolipids is observed along organelles of the secretory pathway with highest levels being found in the plant plasma membrane. The two lipid classes define the thickness and rigidity of membranes which are critical features in membrane organisation (Casares et al., 2019).

(12)

6

Table 1. Plant lipid classes, lipid composition, and abbreviations

Lipid classes Lipid composition and abbreviations

Sphingolipids

Long-chain bases (LCBs) Ceramides (Cers)

Glycosylceramides (GlcCers)

Glycosyl inositiolphosphorylceramides (GIPCs)

Sterol lipids

Free sterols

Steryl glycosides (SG)

Acylated steryl glycosides (ASG) Steryl esters (SE)

Glycerolipids

Polar phosphoglycerolipids

Phosphatidylcholines (PC) Phosphatidylethanolamines (PE) Phosphatidylglycerols (PG) Phosphatidylinositols (PI) Phosphatidylserines (PS) Phosphatidic acid (PA)

Lysoglycerophospholipids (LGPL)

Polar glycoglycerolipids

Monogalactosyldiacylglycerols (MGDG) Digalactosyldiacylglycerols (DGDG) Sulfoquinovosyldiacylglycerols (SQDG) Diacylglyceryltrimethylhomo-Ser/-A (DGTS/A) Lysoglyceroglycolipids (LGGL)

Neutral glycerolipids Diacylglycerols (DAG) Triacylglycerols (TAG)

The plant PM has a specific lipid composition which is similar to animal PM lipid composition. Sterols represent 20–50 mol %; sphingolipids, 5–40 mol %; and phospholipids, 10–60 mol % of plant PM lipids (van Hooren & Munnik, 2017). The lipid constitution varies not only between different plant species but also between different tissue types (Sperling et al., 2005; Markham et al., 2006; Resemann, 2018). Furthermore, the composition is dynamically adjusted when plants are exposed to different biotic and abiotic stresses (Uemura et al., 1995; Nagano et al., 2014). As in animal systems, lipid asymmetry is proposed between the two monolayers of plant PM. DGDG and 60 % of the phospholipids are described to be located in the inner PM leaflet, while glycosylceramides (GlcCers), glycosyl inositolphosphorylceramides (GIPCs), and sterols are mainly found in the outer leaflet (Tjellström et al., 2010; Cacas et al., 2016).

The main phytosterol species are campesterol, ß-sitosterol, and stigmasterol (Furt et al., 2011). The plant PM contains free phytosterols as well as the conjugated phytosterol forms steryl glycosides (SG) and acyl steryl glycosides (ASG) (Furt et al., 2010).

Sphingomyelin is not detected in plants. However, plants contain GIPCs which have varying head group compositions depending on the plant species (Buré et al., 2011; Cacas et al., 2013). The plant equivalent of mammalian glycosphingolipids are GlcCers. The major phospholipids in the plant PM are PC and PE. Phosphatidylglycerol (PG), phosphatidic acid

(13)

7 (PA), PS, PI, and phosphatidylinositol-phosphates (PIPs) are low abundant phospholipids (Mamode Cassim et al., 2019).

The large diversity of the plant plasma membrane lipid composition has raised the question of the involvement of different lipid species in PM organisation. Similar to animal membranes, DRM fractions of plant PMs have been found to be enriched in sterols and sphingolipids (Mongrand et al., 2004; Borner et al., 2005; Minami et al., 2009; Moscatelli et al., 2015). Like the mammalian cholesterol, phytosterols have been described to be able to induce Lo phase formation (Roche et al., 2008; Gerbeau-Pissot et al., 2014). Recent studies on model membranes such as giant unilamellar vesicles (GUVs) and giant vesicles of native PMs (GVPMs) that mimic plant lipid mixtures reported varying abilities of different phytosterols to order membranes (Grosjean et al., 2015; Grosjean et al., 2018). Especially campesterol appears to strongly promote ordered domain formation. Interactions with the highly glycosylated GIPCs were described to enhance the ordering effect of campesterol (Grosjean et al., 2015). In addition to that, immunogold electron microscopy in tobacco (Nicotiana tabacum) PM showed clustering of GIPCs in 35 nm diameter membrane domains (Cacas et al., 2016). Other membrane sub compartments were also found to be enriched in phosphatidylinositol 4,5-bisphosphate (PIP2) (Furt et al., 2010). Localisation studies on membrane proteins further revealed clustering of different proteins in distinct membrane domains (Raffaele et al., 2009; Jarsch et al., 2014; Noirot et al., 2014).

These previous reports hint at lateral heterogeneity in the plant PM that is similar to that observed in animal cells. Membrane domains with different lipid and protein compositions and hence varying biophysical properties appear to co-exist in the plant PM.

The contribution of individual lipid species to membrane organisation is a current research subject. Combining proteomics, lipidomics, and different imaging techniques with the study of mutant plants that are disturbed in PM organisation has greatly helped in understanding membrane dynamics (Grison et al., 2015; Grosjean et al., 2015; Grosjean et al., 2018).

However, the highly complex and dynamic nature of biological membranes poses major challenges to study in vivo mechanisms of plant membrane organisation and leave many unanswered questions that have to be addressed in the future.

2.4 The enigma of the plant sphingolipids

As described, sphingolipids are ubiquitous and essential membrane components in eukaryotes that play major roles in PM organisation. They were first described as lipid components of the brain tissue in the late nineteenth century by Johann Ludwig Wilhelm Thudichum (Thudichum, 1884). It is claimed that he named the newly discovered lipid class

‘sphingolipid’ in allusion to the Sphinx, a creature from Greek mythology. The unique structure of sphingolipids was equally cryptic to scientists at the time of their discovery as

(14)

8

the riddles of the Sphinx. Nowadays, the structure and function of sphingolipids in mammalian cells is well investigated, mainly because of their association with severe metabolic and nervous disorders, some of them known as sphingolipidoses. Lipid storage disorders are genetically inherited and affect various organs and the nervous system. Other disorders that affect sphingolipid metabolism include autoimmune diseases. Examples for sphingolipid-associated diseases are Tay-Sachs disease, Niemann-Pick disease, and Guillain-Barré syndrome (Yu & Ariga, 1998; Sandhoff & Harzer, 2013).

In plants, however, sphingolipids have been an overlooked lipid class for many years. Over the past three decades plant sphingolipids have been associated with multiple essential cellular processes and therefore attracted more attention by the plant science community. Plant sphingolipids may account for up to 10 % of total lipids from plant tissues (Lynch & Dunn, 2004). The structural diversity and complexity of plant sphingolipids requires powerful analytical tools. Up to 200 molecular species in the plant sphingolipidome have been described using advanced mass spectrometric approaches (Markham et al., 2006; Markham & Jaworski, 2007; Cacas et al., 2013). Several plant enzymes involved in sphingolipid biosynthesis have been identified by sequence similarity to characterised enzymes in the baker’s yeast Saccharomyces cerevisiae. Mutants that are disturbed in different sphingolipid enzyme activities have been generated and showed major physiological and metabolic phenotypes (Luttgeharm et al., 2016). The combination of analytical approaches, in vitro enzyme assays, and investigations of plant mutants that are compromised in sphingolipid metabolism have been valuable tools in expanding our understanding of plant sphingolipid structure, metabolism, and function.

2.5 Sphingolipid biosynthesis and structure

Sphingolipids are amphipathic compounds. Their hydrophobic backbone includes an amino alcohol, referred to as long-chain base (LCB). LCBs are the characteristic core of sphingolipids that identify them as a distinct lipid class. LCBs may be connected to a fatty acid moiety. The resulting product is referred to as ceramide, which is the hydrophobic component of sphingolipids. More complex sphingolipid classes are formed through the conjugation of hydrophilic polar head groups to the LCB moiety of the ceramide backbone.

The polar head groups of sphingolipids largely differ between animal, yeast, and plant cells.

Plant sphingolipids are categorised into the following four classes: LCBs, ceramides, GlcCers, and GIPCs (Fig. 1). LCBs and ceramides are minor sphingolipid compounds that constitute 0.5 % and 2 % of the total sphingolipid content in Arabidopsis thaliana leaf extract, respectively (Markham et al., 2006). GlcCers and GIPCs are the most abundant plant sphingolipids and represent 34 % and 64 % of the total sphingolipid content in A. thaliana leaf extract, respectively (Markham et al., 2006).

(15)

9

Fig. 1. Plant sphingolipids are divided into four main classes. Plant sphingolipids are categorised into long- chain bases (LCBs), ceramides, glycosylceramdies (GlcCers), and glycosyl inositolphosphorylceramides (GIPCs). The simplest sphingolipid compound, the LCB, is an amino alcohol. The LCB moiety may be connected via its amino group to a fatty acid moiety. The resulting product is called ceramide. Addition of a glucose moiety to the ceramide backbone results in the formation of GlcCers. Addition of an inositolphosphate and subsequent glycosylation results in the formation of GIPCs, whereby the number of added sugar groups may vary. The overview represents the four plant sphingolipid classes without detailed structural modifications on the ceramide backbone. Modified from (Lynch and Dunn, 2004).

The following description of sphingolipid biosynthesis in plants is based on findings from A. thaliana. Sphingolipids are mainly synthesised via the de novo pathway that is acyl- coenzyme A (CoA) dependent (Fig. 2). Sphingolipid biosynthetic enzymes are located in the membrane of the ER. The de novo pathway is initiated by the condensation of serine and palmitoyl-CoA. The reaction is catalysed by the serine palmitoyltransferase and results in the formation of the intermediate 3-ketosphinganine (Chen et al., 2006; Dietrich et al., 2008; Teng et al., 2008). The enzyme 3-ketosphinganine reductase catalyses the reduction of 3-ketosphinganine to the simplest sphingolipid compound: the LCB sphinganine (Chao et al., 2011). Sphinganine is also referred to as dihydrosphingosine, a dihydroxy LCB, or in short d18:0 (Fig. 2). As the name indicates, d18:0 has a chain length of 18 carbon atoms and contains two hydroxyl groups at the C-1 and C-3 positions. The two hydroxyl groups derive from the serine and palmitoyl-CoA precursors. Different modifications are introduced to the LCB moiety that define its downstream metabolic fate. A third hydroxyl group may be introduced to the C-4 position by an LCB C-4 hydroxylase (Sperling et al., 2001; Chen et al., 2008). The resulting LCB is referred to as phytosphinganine, a trihydroxy LCB, or in short t18:0 (Fig. 2). Most LCB moieties of plant sphingolipids are trihydroxylated (Markham et al., 2006). Double bonds can be introduced to the LCB moiety by two distinct classes of

(16)

10

LCB desaturases. Double bonds may be inserted between the C-4 and C-5 position, designated as Δ4, and between the C-8 and C-9 position, designated as Δ8 (Sperling et al., 1998; Ryan et al., 2007; Michaelson et al., 2009). The most common LCB moiety found in A. thaliana and other plants is trihydroxylated with a double bond in Δ8 position, t18:1 (Fig.

2) (Markham et al., 2006). While the Δ4 double bond is only inserted in trans (E) configuration, the Δ8 double bond can be inserted in either cis (Z) or trans (E) configuration.

The prevalence of the two Δ8 configuration states largely varies between different plant species and tissue types and may change when plants are exposed to external stresses (Markham et al., 2006; Sato et al., 2019).

N-acylation of LCBs is catalysed by ceramide synthases and results in the formation of ceramides (Fig. 2). In plants, LCBs may be connected to fatty acids with chain lengths varying from 16 to 26 carbon atoms. Fatty acids with chain lengths of 16 or 18 carbons (C16, C18) are called long-chain fatty acids (LCFAs), while fatty acids with chain lengths longer than 18 carbons (≥ 20C) are called very long-chain fatty acids (VLCFAs). In A. thaliana distinct ceramide synthases have been described that have different substrate preferences. The class I ceramide synthase generates ceramides with dihydroxy LCBs and LCFAs, while the class II ceramide synthases prefer trihydroxy LCBs and VLCFAs (Markham et al., 2011; Ternes et al., 2011a) (Fig. 2). LCBs and ceramides may also be phosphorylated at the C-1 position of the LCB moiety by the action of LCB and ceramide kinases and are subsequently referred to as LCB phosphates (LCB-Ps) and ceramide phosphates, respectively (Liang et al., 2003; Imai & Nishiura, 2005; Worrall et al., 2008).

Structural modifications may also be introduced to the fatty acid moiety of ceramides. The acyl chain may be hydroxylated at the C-2 or ‘α’ position through the activity of a fatty acid hydroxylase. Ceramides with α-hydroxylated fatty acid moieties are often termed hydroxyceramides. In the nomenclature, a saturated, α-hydroxylated fatty acid moiety with a 24-carbon chain can be called h24:0. If the fatty acid moiety is not hydroxylated, it is often called c24:0. The fatty acid moiety may also carry a cis double bond in n-9 position (Imai et al., 2000).

Ceramides are the precursor molecules for the more complex sphingolipid classes GlcCers and GIPCs. The formation of GlcCers and GIPCs are alternative routes within sphingolipid metabolism (Fig. 2). The second most abundant plant sphingolipid class, GlcCer, is generated by the attachment of a hexose moiety, mostly glucose and sometimes mannose, to the C-1 of the LCB moiety. The transfer of a sugar moiety from uridine diphosphate-glucose (UDP-Glc) is catalysed by a glucosylceramide synthase (Leipelt et al., 2001; Melser et al., 2010; Msanne et al., 2015). The hexose is connected to the ceramide backbone by a 1,4-glycosidic linkage (Leipelt et al., 2001) (Fig. 2).

(17)

11

Fig. 2. Abbreviated de novo sphingolipid biosynthesis in A. thaliana. The majority of reactions within sphingolipid biosynthesis takes place in the endoplasmic reticulum (ER). During the initial steps, the simplest sphingolipid compound, the long-chain base (LCB) sphinganine (d18:0), is formed. The d18:0 LCB is subsequently applied to modifications such as Δ4-desaturation and C-4 hydroxylation. N-acylation of the LCB moiety results in ceramide formation. Depending on the structural features of the LCB, different ceramide synthases, class I or class II, are active that connect dihydroxy LCBs either with long-chain fatty acids (LCFAs), or trihydroxy LCBs with very long-chain fatty acids (VLCFAs). The ceramide backbone may subsequently be modified by additional Δ8-desaturation of the LCB moiety, fatty acid α-hydroxylation, or fatty acid n-9 desaturation. The combination of structurally different LCB and fatty acid moieties results in a large variety of ceramide species that are subsequently channelled into the glucosylceramide (GlcCer) or glycosyl inositolphosphorylceramide (GIPC) formation. The demonstrated pathway is an abbreviated version of sphingolipid biosynthesis, not including reactions such as phosphorylation, de-phosphorylation or breakdown of complex sphingolipids. Abbreviations are as follows: CoA: Coenzyme A; GDP-Man: Guanosine Diphosphate Mannose; GINT1: Glucosamine Inositolphosphorylceramide Synthase; GMT: GIPC Mannosyl Transferase;

GONST1: GDP-Mannose Transporter; IPCS: Inositolphosphorylceramide Synthase; IPUT:

Inositolphosphorylceramide Glucuronosyl Transferase; UDP-Glc: Uridine Diphosphate Glucose.

In contrast to the previous steps in sphingolipid biosynthesis that happen in the ER, modification of the most abundant plant sphingolipid class, GIPC, happens in the Golgi apparatus (Wang et al., 2008). GIPC synthetic enzymes reside in the Golgi membrane and ceramide substrates are therefore exported from the ER and transported to the Golgi apparatus for further processing. Inositolphosphorylceramide synthases transfer an inositolphosphate head group from PI to the ceramide backbone (Wang et al., 2008; Mina et al., 2010). Subsequent glycosylation steps can add up to seven additional sugar residues to the inositolphosphorylceramide head group, leading to a variety of GIPC species with different head group compositions (Mortimer et al., 2013; Rennie et al., 2014; Fang et al.,

(18)

12

2016; Tartaglio et al., 2017; Ishikawa et al., 2018). The first sugar moiety linked to the inositolphosphorylceramide backbone is usually glucuronic acid. The following sugar moieties may be hexosamine, N-acetylhexosamine, or a variety of different pentoses and hexoses. If only one sugar moiety is attached to the glucuronic acid, the GIPCs are called series A GIPCs. If two sugar moieties are attached, GIPCs are termed series B GIPCs. The glycan head group composition varies between different plant species and tissue types (Buré et al., 2011; Cacas et al., 2013).

The introduced modifications in the ceramide backbone including the hydroxylation status (Fig. 3A), the number and position of inserted double bonds (Fig. 3B), the composition of polar head groups (Fig. 3C), and the variation of the acyl chain length (Fig.

3D) are the main causes for the diversity found among plant sphingolipids. The structural features of individual sphingolipid species offer them an array of different biophysical properties including size, charge, or polarity and are likely key to their varied physiological functions.

Fig. 3. Structural modifications on the ceramide backbone broadens variety of A. thaliana sphingolipids.

Structural modifications on the ceramide backbone include (A) hydroxylation of the LCB and the fatty acid moieties, (B) desaturation of the LCB and the fatty acid moieties, (C) the composition of the polar head group, and (D) the chain length of the fatty acid moiety. (A) Hydroxylation can happen on the C-4 position of the LCB moiety or on the C-2 or α position of the fatty acid moiety. (B) Double bonds may be introduced at the Δ4 and Δ8 position of the LCB and at the n-9 position of the fatty acid moiety. (C) Different polar head groups (designated as R) such as glucose (Glc) or inositolphosphate and additional sugar residues (IPGlc) may be added to the C-1 of the LCB moiety. (D) In plants, the fatty acid chain length varies from 16 to 26 carbon atoms.

Modified from (Berkey et al., 2012).

Sphingolipids are involved in various signal transduction processes both during plant development as well as in immune responses. Especially LCBs and ceramides, which are believed to be second messengers in signalling cascades, are considered essential for the establishment of adaptive responses. Minor amounts of LCBs and ceramides may be provided by the breakdown of more complex sphingolipids which enables their re-entry into synthetic pathways. This process is referred to as salvage pathway. Different degradation

(19)

13 enzymes are active during the salvage pathway, including glucosylceramidases (Dai et al., 2020), various ceramidases (Chen et al., 2015; Li et al., 2015; Wu et al., 2015; Zienkiewicz et al., 2020) and an LCB-P lyase (Tsegaye et al., 2007; Nishikawa et al., 2008). In contrast to the de novo biosynthesis pathway, the salvage pathway is fatty acid and not acyl-CoA dependent. To ensure fast responses of plants to their environment, conversion of sphingolipid compounds through anabolic and catabolic enzyme reactions has to adapt quickly.

2.6 Plant sphingolipids have diverse physiological and metabolic functions Sphingolipids have roles as structural elements in membranes and as bioactive molecules during signal transduction (Greenberg et al., 2000; Coursol et al., 2003; Markham et al., 2006; Shi et al., 2007). Disruption of sphingolipid metabolism causes severe defects in essential cellular processes such as development and the plant’s ability to respond to external stresses (Chen et al., 2008; Msanne et al., 2015; Gonzalez-Solis et al., 2020).

Alterations in sphingolipid structure may influence the overall biophysical properties of membrane domains. Furthermore, disruption of sphingolipid metabolism may interfere with signalling cascades during essential cellular processes. This shows that an imbalance of sphingolipid homeostasis has drastic and harmful effects on plant viability (Abbas et al., 1994; Chen et al., 2008; König et al., 2012; Msanne et al., 2015; Gonzalez Solis et al., 2020;

Zienkiewicz et al., 2020). Therefore, the conversion of sphingolipids must be controlled in a dynamic manner to avoid an unusual accumulation of certain sphingolipid compounds that negatively affect plant viability (Abbas et al., 1994; Liang et al., 2003; Shi et al., 2007; Chen et al., 2008).

Because of their distinct structural features, the four plant sphingolipid classes and even certain sphingolipid species have been ascribed to different physiological functions.

The less abundant LCBs and ceramides appear to mostly act as bioactive mediators of cellular functions. LCBs, ceramides and their phosphorylated forms seem to be antagonistic partners in these processes. Especially the balance between LCBs, ceramides and their phosphorylated counterparts are important factors in regulating physiological processes.

The activity of sphingolipid kinases and lyases controls the ratio of the free and phosphorylated forms (Liang et al., 2003; Imai & Nishiura, 2005; Worrall et al., 2008). LCBs and ceramides both have been reported as triggers of programmed cell death (PCD) (Greenberg et al., 2000; Liang et al., 2003; Shi et al., 2007; Alden et al., 2011). First indications for the involvement of LCBs and ceramides in PCD induction were observed during studies with fungal-derived sphingosine analogues. The mycotoxins from Alternaria alternata f. sp. lycopersici (AAL) and fumonisin B1 (FB1) from Fusarium species are able to elicit PCD in plants. The two mycotoxins have structural similarity to sphingosine and

(20)

14

therefore act by competitively inhibiting ceramide synthase activity. Blockage of ceramide synthesis resulted in elevated levels of LCBs (Abbas et al., 1994; Stone et al., 2000).

Following studies reported the specific inhibition of the class II ceramide synthase LOH1 by FB1, which in turn resulted in elevated levels of dihydroxy ceramides with LCFAs (Markham et al., 2011; Molino et al., 2014). LCBs and ceramides were later also directly shown to elicit PCD symptoms. Exogenous application of LCBs to A. thaliana leaves resulted in the induction of reactive oxygen species (ROS) dependent PCD (Shi et al., 2007). Conversely, simultaneous application of LCBs and LCB-Ps suppressed the onset of PCD, which indicated that LCBs and LCB-Ps appear to counteract with each other (Shi et al., 2007;

Alden et al., 2011). Similarly, investigation of the ceramide kinase mutant acd5 revealed accumulation of ceramides, which was accompanied by PCD symptoms (Greenberg et al., 2000; Liang et al., 2003). Analogous to the ratio of LCBs to LCB-Ps, the balance of ceramides and phosphorylated ceramides was also observed to be a critical factor in cell death induction (Liang et al., 2003). In general, ceramides and LCBs appear to play crucial roles in plant resistance to pathogens. Elevated levels of LCBs and ceramides were not only associated with the onset of PCD but also with upregulation of defence-related genes and higher levels of certain phytohormones. Especially a correlation of sphingolipid metabolism and phytohormone signalling appears to be a key factor in mediating the plant immune response. The fatty acid hydroxylase mutant, fah1 fah2, was shown to accumulate LCBs and ceramides, had constitutive PR1 and PR2 expression, and higher salicylic acid (SA) levels (König et al., 2012). More recently, A. thaliana mutants disrupted in neutral ceramidase activities, ncer1 and ncer2, accumulated jasmonic acid-isoleucine (JA-Ile) and SA, respectively (Zienkiewicz et al., 2020). ncer1 plants had higher levels of hydroxyceramides, which was associated with early leaf senescence (developmentally controlled PCD), while ncer2 plants had higher levels of t18:0 LCBs, which was associated with defence-related cell death (pathogen-triggered PCD). The differing cell death symptoms in the two independent neutral ceramidase knockouts indicate that elevated levels of LCBs and ceramides may elicit different downstream signalling cascades that include either JA or SA pathways. In addition to their role in plant immune responses, LCB- Ps were also associated with abscisic acid (ABA) dependent guard cell closure (Ng et al., 2001; Coursol et al., 2003). Mutants disrupted in their sphingosine kinase (SphK1) activity were less sensitive to ABA-promoted stomatal closure. ABA is proposed to activate SphK1 which in turn caused an increase in LCB/LCB-P ratio. The signalling cascade affected cytosolic ion levels and hence opened ion channels that in turn caused turgor reduction of the guard cells, resulting in stomatal closure. Cumulative findings concerning LCB and ceramide signalling in plant cells indicate the participation of different LCB and ceramide species in response to biotic and abiotic stresses.

(21)

15 The more abundant complex plant sphingolipid classes GlcCers and GIPCs are ubiquitous structural elements of the plant PM and of endomembrane systems. They have been detected as components of intracellular membranes, including ER, Golgi apparatus, tonoplast, and endosomes (Moreau et al., 1998; Mongrand et al., 2004; Sperling et al., 2005; Bayer et al., 2014). A sphingolipid gradient is observed along the secretory pathway with highest sphingolipid levels found in the PM. GlcCers and GIPCs compose around 5- 10 % and 40 % of all plant PM lipids, respectively, and are considered to be enriched in the outer leaflet (Tjellström et al., 2010; Cacas et al., 2016). The relative abundances of GlcCer and GIPCs in the plant PM likely contribute to adaptive processes towards biotic and abiotic stresses. For instance, the ratio of GlcCers to GIPCs in the plant PM has been associated with membrane adjustments in response to cold stress. Nagano et al. (2014) reported an increase in GIPC levels and a decrease in GlcCer levels in A. thaliana cold acclimation(Nagano et al., 2014). Although both complex sphingolipid classes are assumed to be enriched in the plasma membrane (Cacas et al., 2016), the two classes are structurally distinguishable in their head group and in their ceramide backbone composition. Therefore, they might have different functions in plant physiology. However, differences in their physiological activities are still poorly understood.

GlcCers are described to be specifically enriched in A. thaliana pollen and floral tissue (Luttgeharm et al., 2015b). GlcCer-deficient mutants cannot develop beyond seedling stage, are defective in organ-specific cell differentiation, have an altered Golgi morphology, and impaired pollen transmission (Msanne et al., 2015). Inhibition of the glucosylceramide synthase, GCS, from A. thaliana with the chemical inhibitor D,L-threo-1-phenyl-2- palmitoylamino-3-morpholino-1-propanol (PDMP) was similarly associated with an altered Golgi morphology (Melser et al., 2010). Cumulative findings indicate a role for GlcCers in Golgi-mediated protein secretion and subsequent vesicle trafficking to the plant PM. The desaturation status of GlcCers was also found to be important in plant response to chilling and freezing. While in chilling-resistant plants the fatty acid moiety of GlcCers was mainly composed of unsaturated α-hydroxylated fatty acids (Cahoon & Lynch, 1991; Imai et al., 1995), GlcCers of chilling sensitive plants did not have those fatty acids (Imai et al., 1995).

GIPCs are considered to be the most abundant plant sphingolipid class, however, the relative abundances of GlcCers and GIPCs may vary depending on the investigated plant species, the tissue type, and the applied external conditions (Sperling et al., 2005;

Markham et al., 2006; Markham & Jaworski, 2007; Luttgeharm et al., 2015b). Due to their complex, highly polar head group compositions, GIPCs have limited solubility in traditionally used extraction solvents. Only recently, extraction methods for plant GIPCs have been optimised and enabled first investigations on this long-overlooked plant sphingolipid class (Buré et al., 2011; Cacas et al., 2013). Their high abundance in the PM, which was recently

(22)

16

described to be around 40 mol % of tobacco PM lipids, puts them into the spotlight as most abundant plant sphingolipids (Cacas et al., 2016; Gronnier et al., 2016). As described in part 2.3, lipid rafts are highly enriched in sterols and sphingolipids. The biggest proportion of sphingolipids found in lipid rafts is represented by GIPCs. GIPCs were found to be enriched in DRM fractions of tobacco Bright Yellow 2 (BY-2) cell cultures (Cacas et al., 2016). Microscopic evidence for GIPC enrichment in certain membrane domains was given by Cacas et al. (2016), who performed immunogold labelling of tobacco PM vesicles.

Subsequent tissue investigation with electron microscopy revealed clustering of highly glycosylated GIPCs in distinct membrane domains (Cacas et al., 2016). GIPCs were further reported to enhance the campesterol-induced ordering effect of membrane domains (Grosjean et al., 2015). Through their structural function in lipid rafts, GIPCs are assumed to be involved in a multitude of PM-associated signal transduction processes. They are described as lipid anchors for GPI-anchored surface proteins (Borner et al., 2005; Lefebvre et al., 2007). Plant GIPCs might also act as cell wall anchors (Voxeur & Fry, 2014).

Moreover, GIPCs were recently also identified as toxin receptors and are known to be involved in salt sensing (Lenarčič et al., 2017; Jiang et al., 2019). In addition to that, purification of plasmodesmata membrane fractions reported a similar lipid composition as described for membrane rafts (Grison et al., 2015). GIPCs are therefore likely involved in plasmodesmal cell-to-cell transport of nutrients and signalling compounds (Yan et al., 2019;

Liu et al., 2020).

2.7 Metabolic routing of sphingolipid intermediates

Apart from their physiological function in plants, structural modifications on the ceramide backbone may also have a role in channelling ceramide substrates into downstream complex sphingolipid synthesis. As mentioned, GlcCer and GIPC formation display two alternative pathways in sphingolipid metabolism (Fig. 2). Especially the hydroxylation and desaturation state of the LCB moiety is considered to be responsible for dictating the metabolic fate of precursor compounds. Previous studies on A. thaliana showed that the trihydroxy LCB, mostly t18:1 with the double bond in Δ8 position, is the most abundant moiety in GIPCs while LCB Δ4-desaturation likely plays a key role in channelling substrates into GlcCer formation (Chen et al., 2008; Michaelson et al., 2009). The t18:1 LCB moiety of GIPCs is mainly found in association with α-hydroxylated VLCFAs, while the d18:2 LCB moiety of A. thaliana pollen GlcCers and of species such as tomato (Solanum lycopersicum) and soybean (Glycine max) is mostly connected with the α-hydroxylated LCFA C16. The channelling function of the C-4 hydroxylation and the LCB Δ4-desaturation is supported by the fact that both reactions happen on the C-4 position of the LCB moiety which makes them mutually exclusive.

(23)

17 The prevalence of certain ceramide modifications in GlcCers and GIPCs appears to have direct effects on the physiological functions of the two complex sphingolipid classes.

The typical ceramide backbone found in GlcCer and GIPC species is usually highly hydroxylated, both on the LCB as well as on the fatty acid side. The hydroxylation status of sphingolipids is considered essential for building up an extensive hydrogen bond network with other membrane components (Slotte, 1999; Mombelli et al., 2003; Slotte, 2016). This is especially important for the interaction with phytosterols during lipid raft formation (Mamode Cassim et al., 2019). A. thaliana has two functionally redundant LCB C-4 hydroxylases, SBH1 and SBH2. The combined activities of both enzymes account for all trihydroxy LCB formation in the plant. Both genes were able to complement the S. cerevisae LCB C-4 hydroxylase knockout sur2Δ (Sperling et al., 2001). Knockout of both hydroxylase encoding genes led to severely dwarfed plants that were likely disturbed in cell elongation and division and that did not reach reproductive maturity (Chen et al., 2008). Additionally, knockout plants showed necrotic cotyledon lesions which were accompanied by the upregulation of defence-related marker genes. Knockout of both genes caused serious alterations in all sphingolipid classes (Chen et al., 2008). The most prominent observation was a drastic accumulation of sphingolipids containing dihydroxy LCB moieties and C16 fatty acid moieties. Furthermore, the most abundant LCB moiety in all sphingolipid classes switched from trihydroxy LCBs to dihydroxy LCBs (Chen et al., 2008). Since sphingolipid content and composition were both affected in the mutant, the phenotype was speculated to derive either from the unusual accumulation of sphingolipids with dihydroxy LCBs and C16 fatty acids or from a global change in the most abundant LCB moiety from trihydroxy to dihydroxy LCBs. Similar to the LCB C-4 hydroxylases, A. thaliana harbours two redundant fatty acid hydroxylases, FAH1 and FAH2 (König et al., 2012). The fah1 fah2 mutant has reduced levels of α-hydroxylated sphingolipids and instead showed elevated levels of sphingolipids with unhydroxylated fatty acid moieties. Furthermore, trihydroxy LCBs and ceramides were enriched five- and ten-fold, respectively, and total GlcCer levels were reduced by 25 % compared to the wild type. These alterations in the sphingolipidome were accompanied by reduced plant size, elevated SA levels, constitutive PR gene expression and an associated increased resistance against the obligate biotrophic pathogen Golovinomyces cichoracearum (König et al., 2012).

Another important feature is the prevalence of VLCFAs in GlcCers and GIPCs.

Longer acyl chains increase the hydrophobicity and the membrane transition from fluid to gel phase state (Pinto et al., 2014; Mamode Cassim et al., 2019). Additionally, the chain length of the fatty acid moiety is considered a crucial feature in the interdigitation and therefore in the connection of the inner and outer PM monolayers (Mamode Cassim et al., 2019). Sphingolipids with VLCFAs are reported to have a crucial role in development

(24)

18

(Markham et al., 2011; Molino et al., 2014). Inhibition of ceramide synthases that are specific for VLCFA substrates caused defects in root-outgrowth. On a subcellular level, defects in VLCFA-containing sphingolipids resulted in impaired membrane trafficking of auxin proteins to the PM (Markham et al., 2011). Molino et al. (2014) also demonstrated altered cell plate formation in plants whose VLCFA incorporating ceramide synthase LOH1 was blocked by the mycotoxin FB1 (Molino et al., 2014). The authors propose a function for VLCFA- containing sphingolipids in lipid bilayer fusion and therefore in vesicle dynamics during development.

Although GlcCer and GIPC architectures share some structural features, the A. thaliana GlcCer pool differs in certain molecular species from the GIPC pool.

A. thaliana GlcCers are enriched in the Δ4,8-diunsaturated, d18:2, LCB moiety compared to GIPCs (Markham et al., 2006). Plants lacking the two LCB Δ8-desaturases, SLD1 and SLD2, have GlcCer levels that are 50 % reduced compared to the wild type and the mutants are more sensitive to cold stress (Chen et al., 2012). It might be that the configuration state of the Δ8 double bond plays a role in shunting t18:1 species into GlcCer or GIPC formation. In contrast to that, knockout mutants of the A. thaliana LCB Δ4- desaturase did not show any obvious phenotypes (Michaelson et al., 2009). In A. thaliana the d18:2 LCB moiety is enriched in pollen and floral tissue (Michaelson et al., 2009). GlcCer levels were also significantly reduced in the LCB Δ4-desaturase knockout plant, indicating that LCB Δ4-desaturation has indeed a channelling function for GlcCer formation (Michaelson et al., 2009). However, pollen and general plant viability was not affected in the A. thaliana mutant plants. While Δ8-desaturation is one of the most abundant LCB modifications found in A. thaliana sphingolipids, LCB Δ4-desaturation appears to not have a significant physiological role in Brassicaceae (Markham et al., 2006). A lipidomics screen covering 21 plants from different lineages identified the prevalence of the LCB double bond position in d18:1 LCB moieties (Islam et al., 2012). They revealed that LCB Δ4-desaturation is most common to non-vascular plants and to the Poales family whereas LCB Δ8- desaturation is most abundant in plants like Brassicaceae. The authors speculated that LCB Δ4-desaturation appears to be more ancient than LCB Δ8-desaturation. Interestingly, in plants like tomato and soybean ceramides with a Δ4,8-diunsaturated LCB moiety and C16 fatty acids are most abundant (Markham et al., 2006). This suggests that the LCB desaturation state was subject to divergent evolution and that LCB Δ4-desaturation likely has a more important physiological role in plants outside the Brassicaceae family. In the filamentous fungus Pichia pastoris loss of LCB Δ4-desaturation has more pronounced metabolic effects resulting in complete abolishment of GlcCers (Michaelson et al., 2009).

Plants that lack all GlcCers are seedling lethal and plants with a disturbed GlcCer formation show defects in cell differentiation and organogenesis, likely due to an impaired intracellular

(25)

19 membrane trafficking (Melser et al., 2010; Melser et al., 2011; Krüger et al., 2013). The structural features of GlcCers therefore appear to have an important role in vesicle dynamics.

GIPCs have a characteristic head group. The glycan residues of the GIPC head group might be important for pathogen perception and cell recognition (Lenarčič et al., 2017). The head group composition is strongly specific for certain plant species and tissue types (Buré et al., 2011; Cacas et al., 2013; Luttgeharm et al., 2015b). Alterations of the glycan head group composition is reported to have severe effects on plant viability.

Knockout of one of the three inositolphosphoceramide synthases, ERH1, resulted in GIPC reduction and accumulation of the ceramide precursor, which was accompanied by the onset of cell death symptoms (Wang et al., 2008). Knockout of subsequent enzymes that catalyse conjugation of different sugar residues resulted in mutants that were either lethal or had severe growth defects (Mortimer et al., 2013; Rennie et al., 2014; Tartaglio et al., 2017).

Taken together, studies on distinct A. thaliana sphingolipid mutants indicate that LCB modifications have a strong influence on the metabolic flux of sphingolipid compounds, and that distinct structural features of GlcCers and GIPCs have important effects on their physiological function. Especially LCB C-4 hydroxylation and LCB Δ4-desaturation appear to be of great importance for the downstream metabolic fate of sphingolipids. In A. thaliana, the channelling of sphingolipid metabolites seems to also be partially controlled by the ratio of cis and trans Δ8 double bonds (Markham et al., 2006; Markham & Jaworski, 2007).

However, the exact channelling process of sphingolipid intermediates in plants is not yet fully elucidated, in part because of the large complexity of the A. thaliana sphingolipidome.

Furthermore, although sphingolipid biosynthesis is broadly conserved among plants of different taxonomic groups, sphingolipid composition differs between plant species and even between different tissues of the same plant (Sperling et al., 2005; Markham et al., 2006; Markham & Jaworski, 2007; Luttgeharm et al., 2015b). This underlines that distinct sphingolipid species have different physiological roles which may be more or less important in certain plants and plant tissues. This also includes a potential divergent evolution of different pathogen interaction systems in plants of different taxonomic groups. A recent study gave a great example for this and showed that GIPCs can act as necrosis and ethylene-inducing peptide 1–like (NLP) toxin receptors in eudicots but not in monocots (Lenarčič et al., 2017). The authors concluded that this host selectivity may be due to the glycan head group composition which is known to be plant species- and tissue-dependent (Buré et al., 2011; Cacas et al., 2013). These observations indicate that sphingolipid metabolism diverged during land plant evolution. However, the functional relevance for this diversification is largely unknown.

(26)

20

2.8 The value of bryophytes in sphingolipid studies

Most findings on plant sphingolipid biosynthesis have been gained from studies on the vascular model A. thaliana (Luttgeharm et al., 2016). While observations made in this commonly used model contributed greatly to our knowledge on plant sphingolipid metabolism, the sole focus on the A. thaliana sphingolipidome may also present some limitations. As the previous paragraph explained, sphingolipid metabolism in plants appears to have diverged during land plant evolution. Therefore, structural modifications such as LCB Δ4-desaturation, which is an essential feature for GlcCer formation in plants like tomato or soybean, is nearly absent in A. thaliana (Markham et al., 2006; Michaelson et al., 2009).

To study the physiological relevance of compounds that are not important for Brassicaceae, it is therefore crucial to study plants beyond A. thaliana (Michaelson et al., 2009; Islam et al., 2012; Markham et al., 2013). Another disadvantage of studying sphingolipid biosynthesis in vascular land plants is their complex body plan. Disruption of sphingolipid genes in vascular plants therefore often results in pleiotropic phenotypes, making it difficult to ascribe distinct sphingolipids to certain physiological functions (Chen et al., 2008; König et al., 2012; Msanne et al., 2015). Our recently acquired knowledge on the importance of plant GIPCs in membrane organisation furthermore calls for novel tools to study in planta membrane dynamics. Previous studies focused on working with model membrane systems of decreasing complexity to decipher the role of individual sphingolipid classes on membrane organisation (Grosjean et al., 2015; Grosjean et al., 2018). While these studies gave undeniably valuable insights of how different sphingolipid classes influence the membrane order in vitro, they unfortunately omit the complex background of biological membranes in native tissues. This includes the presence of integral or peripheral membrane proteins and intercellular communication. The study of plants with simpler tissue types and a sphingolipidome of lower complexity might therefore be key in advancing our knowledge on the sphingolipid function in plants.

The study of different plant lineages displays a valuable tool in understanding the evolution of physiological processes. Around 400 million years ago the first plants conquered terrestrial environments as habitat. A sister lineage of vascular plants are the bryophytes that comprise three groups of non-vascular land plants: mosses, liverworts, and hornworts (Hedges, 2002). Emerging model organisms from this group are the liverwort Marchantia polymoropha and the moss Physcomitrella patens. Mosses share the same essential metabolic and physiological processes with vascular plants. However, their body plans are much simpler than the highly complex vascular tissues. P. patens has gained increased attention in plant research over the past three decades due to a plethora of advantageous properties (Cove, 2005; Cove et al., 2006; Rensing et al., 2008). In 2008 the

(27)

21 P. patens genome was completely sequenced (Rensing et al., 2008). Since then an advanced molecular toolkit for gene editing of the plant has been developed. The moss has a haplodiplontic life cycle with the alternation of two generations: the haploid gametophyte and the diploid sporophyte. The dominant haploid gametophytic phase of the moss is easily accessible for genetic manipulation. Homologous recombination has for a long time been the method of choice for targeted gene disruption in P. patens (Schaefer & Zrÿd, 1997).

More recently, the use of the CRISPR-Cas9 system has been established in the moss, enabling simultaneous disruption of numerous genes (Lopez-Obando et al., 2016;

Collonnier et al., 2017).

Compared to vascular plants, the architectural design of P.patens is relatively simple. The gametophyte consists of two developmental stages: the initially formed protonema and the shoot-like gametophore (Prigge & Bezanilla, 2010) (Fig. 4). The protonema is a two-dimensional network of filamentous cells. Two cell types compose the protonema. The first cells to be generated are the assimilatory chloronema that harbour numerous chloroplasts. Chloronema cells gradually differentiate into the foraging caulonema cells, which are much longer and grow faster than chloronema cells (Fig. 4). In dark conditions, only caulonema cells are able to grow against the gravity vector and are then referred to as skotonema cells (Cove et al., 1978; Rensing et al., 2020). The filamentous cells grow via polarized tip growth and side-branching is initiated at subapical cells.

Fig. 4. P. patens life cycle. The haplodiplontic P. patens life cycle is dominated by the haploid gametophyte that consists of spores, protonema, and gametophore. It starts with a single spore that germinates into the filamentous protonema. The protonema is composed of two cell types: the chloronema and the caulonema cells.

Buddings that emerge from the protonema grow out into the gametophore. The gametophore has leaf-like phyllids and root-like rhizoids. The reproductive organs, the female archegonia and the male antheridia, are located at the tip of the gametophore. After the egg inside the archegonium is fertilised by spermatozoids, the zygote matures into the spore capsule. The diploid sporophyte consists of a spore capsule and a short seta.

Modified from (Rensing et al., 2020).

Once the protonema has ensured proper establishment of the plant, the development of the adult stage, the gametophore, is initiated (Fig. 4). The gametophore is a three-dimensional structure that is most similar to the shoot of vascular plants. However, the individual organs

(28)

22

of the plant have a much simpler architecture than the ones of vascular land plants. The gametophore shoot carries leaflets that are only one cell layer thick. They also have root- like structures, filamentous rhizoids, which anchor the shoot to the ground and ensure nutrient supply (Fig 4). Sexual reproduction of the moss is dependent on autumn-like environmental conditions. Colder temperatures induce the formation of the sexual organs, the gametangia, at the tip of the gametophore (Hohe et al., 2002) (Fig 4). Subsequent submersion with water enables spermatozoids to swim from the male antheridia to the female archegonia and to fertilise the egg inside the female organ. The zygote matures into the sporophyte, the only diploid phase of the life cycle, which consists of a spore capsule and a short seta (Landberg et al., 2013; Hiss et al., 2017). Bursting of the spore capsule results in the release of haploid spores that initiate a new life cycle (Engel, 1968).

In addition to sexual reproduction P. patens can also be propagated vegetatively.

Disruption of gametophytic tissue results in a high regeneration rate. This might be especially interesting for studying mutants disturbed in sphingolipid metabolism since many A. thaliana sphingolipid mutants are unable to reach reproductive maturity. Disturbed sexual reproduction is often associated with embryo lethal phenotypes. True A. thaliana sphingolipid knockout plants are therefore often not accessible for detailed phenotype characterisation, forcing scientists to instead work on knockdown plants. The life cycle of P. patens can easily be completed under laboratory conditions making each developmental stage easily accessible for thorough phenotypic investigations.

All these aspects have put P. patens into the spotlight for developmental and evolutionary genetic studies. Because of its anatomical simplicity, the yeast S. cerevisiae has been a valuable model in investigating the sphingolipid pathway (Riezman, 2006;

Dickson, 2010). The same reasoning may be applied in using the bryophyte P. patens as plant model with a simple non-vascular anatomy. In terms of studying sphingolipid biosynthesis in plants, discoveries of individual reactions in P. patens sphingolipid biosynthesis may allow us to take a step back from the thoroughly investigated A. thaliana sphingolipidome and put new findings on plant sphingolipid biosynthesis into an evolutionary context. This might help us to understand the divergence of sphingolipid metabolism in different land plant lineages.

2.9 The P. patens sphingolipidome

A recent study conducted a global lipid analysis on P. patens protonema (Resemann, 2018). For the lipidomics screen a multiple reaction monitoring (MRM)-based ultra- performance liquid chromatography (UPLC) coupled with nanoelectrospray ionisation (nanoESI) and triple quadrupole tandem mass spectrometry (MS/MS) approach was applied that was adjusted from a screen for A. thaliana lipids (Tarazona et al., 2015). The

Sphingolipid biosynthesis in Physcomitrella patens

Physcomitrella patens

Table of Contents

1 Abstract

2 Introduction